Wine industry implementation of the NZSEE guidance
on the seismic design of liquid storage tanks
E.V. Au, A.F. Walker & W.J. Lomax
Structex, Consulting Engineers
2015 NZSEE
Conference
ABSTRACT: The New Zealand Society for Earthquake Engineering (NZSEE 2009)
Seismic design of liquid storage tanks guidelines provide excellent procedures for
analysing the seismic performance of tanks and determining design actions. However
given the wide scope of the document, there is a lack of specific guidance that deals with
the subtle nuances of the New Zealand wine industry and the wine tank configurations
which are commonly used. The recent 2013 Seddon earthquake sequence resulted in
significant damage to wine tanks and associated infrastructure, highlighting many
deficient design details resulting from poor design. There is an apparent knowledge gap
between the use of the NZSEE guidance and its implementation into the wine industry.
This paper summarises the types of damage observed and provides indications of likely
causes. This information goes on to outline loadpaths and mechanisms which should be
checked to provide a robust wine tank system, introducing some aspects which are not
explicitly covered by the NZSEE guidance.
1 INTRODUCTION
The seismic assessment and design of liquid storage tanks is not well covered in the New Zealand
loadings and materials standards. In lieu, the New Zealand Society for Earthquake Engineering
(NZSEE 2009) guidance document provides excellent procedures for determining design actions and
analysing the performance of tanks.
Whilst the NZSEE document is assumed to be utilised in the design of tanks for the New Zealand wine
industry, there is a lack of specific guidance that work with the nuances of the wine industry and the
common tank configurations which are used. The recent 2013 Seddon earthquake sequence resulted in
significant damage to wine tanks and associated infrastructure, highlighting the inadequacy of many
design details common in the industry.
The New Zealand wine industry contributes over $1.5 billion to the country’s GDP and supports over
16,500 full-time equivalent jobs (NZIER 2009). Thus the seismic risk posed by poor tank design is not
something that should be ignored.
This paper begins by briefly describing aspects of wine tank design that the NZSEE (2009) document
covers well. This is followed by a summary of damage observed to wine tanks in Marlborough
following the 2013 Seddon earthquake and the common design deficiencies which caused them. The
paper then finishes by describing the load-paths and mechanisms that are not explicitly covered by the
NZSEE document, which should be considered in the seismic design of tanks for the New Zealand
wine industry.
2 COVERAGE OF NZSEE (2009) GUIDANCE DOCUMENT
The scope of NZSEE (2009) guidance is rather broad and appears to cover a wide range of tanks and
configurations, including steel and concrete tanks, vertical or horizontal cylindrical tanks, rectangular
tanks, elevated tanks and tanks on grade, anchored and unanchored tanks, sealed tanks and tanks
where the contents can slosh.
Paper Number O-72
Tanks used for wine storage in New Zealand form a much smaller subset. Wine tanks are generally
sealed vertical cylindrical tanks, with height to radius ratios on the order of 3-5, and constructed from
2-6mm thick stainless steel. Smaller 5,000-60,000 litre tanks are typically supported on mild steel
base-frames, with or without diagonal bracing (Figure 1), and with the frame feet either fixed to the
slab or left unbolted. Larger 40,000-270,000 litre tanks are typically mounted on concrete plinths, and
usually anchored to the slab. Methods used to anchor plinth-mounted tanks are wide ranging, and
include shear bolts anchored into the plinth through the tank skirt (Figure 2a), or necked and unnecked tension bolts epoxied into the slab that are connected to the tank via brackets/chairs, which
themselves vary in configuration (Figure 2b-f).
(a) Diagonally-braced base-frame
(b) Un-braced base-frame
Figure 1. Examples of base-frame mounted tanks.
For vertical cylindrical steel wine tanks, the NZSEE (2009) document provides excellent guidance for
the determining seismic design actions in accordance with NZS1170.5. It considers the convective
(sloshing) mode, rigid and flexible impulsive modes, and provides design charts and worked examples
for choosing the respective modal periods, masses, mass heights and damping. Guidance is given to
the designer on appropriate ductility factors depending on the tank configuration, anchorage and
critical failure mechanism. Prescriptive guidance is given for combining said modes and resolving the
actions into base shear and overturning moment.
As wine tanks are usually sealed, the convective mode is usually constrained and the total wine mass
acts in the impulsive mode. Whilst the guidance document mentions this, little explanation is given for
how modal mass heights are to be increased to account for the full mass acting in the impulsive mode.
This is left to the designer to interpret and apply in a rational manner.
At a tank design level, good guidance is given for determining and checking hoop and bending
stresses in the tank wall, and checking vertical wall stresses against diamond-shaped and elephant’s
foot (elastic-plastic) buckling. For wine tanks which are sealed, a method is provided for determining
the impact loads on the roof cone from the constrained sloshing. However, no method is given for
assessing the strength of the roof cone for resisting such loads. For this, the designer will need to refer
elsewhere, such as API-620 (2008), or undertake software-assisted 3-dimensional shell analysis.
661
For the design of hold-down anchors, equations are provided to calculate tensile demands based on an
anchor force distribution which is dependent on the chosen ductility. Beyond this, the designer is left
to design the anchor and connection to the tank based on the computed demand and applied
overstrength (if applicable).
(a) Shear bolt anchors
(b) Chair mounted un-necked tension anchors
(c) Chair mounted un-necked tension anchor
(d) Chair mounted un-necked tension anchor
(e) Chair mounted necked tension anchors
(f) Chair mounted necked tension anchors
Figure 2. Examples of plinth-mounted tank anchorages.
662
For elevated tanks mounted on base-frames, the NZSEE (2009) document does not make much
comment on the design of the supporting frame. Once seismic design actions have been obtained, a
frame should be able to be designed in accordance with good engineering practice and the New
Zealand steel structures standard, NZS3404:1997. In spite of this, deficient base frame details have
managed to emerge, as will be seen in the following section.
For the design of foundations, the guidance document conceptually covers site investigation, strength
reduction factors for soil, and special ground cases (e.g. slope stability and liquefaction). Although not
in-depth, much of the expectation is the same as for normal building structures.
Given the wide scope of the document, it is understandable that it would not cover all the subtle details
which are unique to the wine industry and the tank configurations which are used. Subtle differences
in liquid-storage tanks which are unique to each industry need to be carefully evaluated using basic
engineering principles and judgement.
3 OBSERVATIONS FROM THE 2013 SEDDON EARTHQUAKES
Damage observed to wine tanks in Marlborough following the 2013 Seddon earthquakes have been
well-documented by others (Morris, Bradley, Walker, Matuschka 2013), but are described again
herein to highlight the importance of detailing on seismic performance, and the subtle details which
need to be considered with wine tanks.
3.1 Damage to plinth-mounted tanks
Typical damage observed to plinth-mounted tanks included:
•
Tensile anchor bolt failure, whether by steel fracture, pull-out through the epoxy, or concrete
cone failure (Figure 3a). Concrete failure modes occur when the anchorage has insufficient
capacity to develop the overstrength of the steel cross section. Steel fracture occurs when the
material has insufficient ductility for the displacement demand, which may occur with
tension-only anchors when an event larger than the chosen design level occurs (described
later in section 5).
•
Shear failure of shear bolts (Figure 3c).
•
Lateral sliding of the tank on top of the plinth, leading to bending of hold-down anchors
(Figure 3d). For older wine tanks, there can be a 20-80mm gap between the tank skirt and
concrete plinth to provide tolerance for the construction method used. As such, the only
available shear mechanism between the tank and plinth is friction between the tank floor and
plinth. When this is exceeded, sliding occurs.
•
“Knuckle-squash” (also referred to as guttering of the knuckle) as depicted in (Figure 3e),
which occurs due to a lack of a complete engineered loadpath. Tension anchors provide holddown to resist overturning. Conversely, there is a compression reaction on the opposite side
which must be resisted. In cases where the tank skirt did not bear on the slab, this loadpath
did not exist and rocking of the tank resulted in deformation of the knuckle. In cases where
hold-down anchors had a nut to the underside of the chair and could take compression,
anchors buckled resulting in milder-deformation of the knuckle.
•
Diamond-shaped or elephant foots’ wall buckling (Figure 3b), which occurs due to
insufficient wall thickness to resist axial wall stresses.
•
Local skirt damage. In the case of Figure 3f, the skirt did not extend to the slab and the
loadpath for the compression reaction was via the hold-down chair. The skirt had insufficient
thickness or support to resist the concentration of stresses, resulting in local buckling around
the chair.
663
(a) Combined concrete cone-pull-out failure
(b) Buckling of tank wall
(c) Steel shear bolt failure
(d) Lateral sliding on top of plinth
(e) Illustration of “knuckle-squash”
(f) Local buckling of skirt
Figure 3. Examples of damage observed to plinth-mounted tanks (Source of figures d-f: Morris et al.
2013).
664
3.2 Damage to base frame-mounted tanks
Typical damage observed to base frame-mounted tanks included:
•
Bending of legs for un-braced frames (Figure 4a). Where frames are un-braced, they are
required to act as moment-resisting frames to resist lateral loads. However, often beam to leg
joints were not detailed to act as fully-rigid connections.
•
Buckling of braces for braced base frames. In these cases, braces were too slender to resist
the applied loads (Figure 4b).
•
Bending of adjustable feet (Figure 4a). Winery floors are often sloped to provide drainage. To
accommodate this, base frame-mounted tanks often have threaded adjustable feet. These
connections were observed to act as weak points leading to failure.
•
Fracture or damage to welded-straps/tags connecting the tank to the base frame. For ease of
construction, often the base frame perimeter did not align with the tank wall, such that tags
ended up being welded to the curved knuckle of the tank. This created an eccentric loadpath
which resulted in fracture of the weld or deformation of the knuckle. In some cases, damage
was sufficient to rupture the knuckle resulting in a loss of contents.
(a) Bending of un-braced leg
(b) Buckling of braces
Figure 4. Examples of damage observed to base frame mounted tanks (Source: Morris et al. 2013).
3.3 Attached catwalks and services
In the New Zealand wine industry, catwalks and piped services are often supported off wine tanks. In
some cases, these secondary structures were rigidly fixed to several tanks, creating an unintentional
loadpath between tanks. Resulting force transfer between tanks often resulted in local damage at
catwalk to tank connections.
4 DESIGNING TANKS FOR THE NEW ZEALAND WINE INDUSTRY
Despite the availability of the NZSEE (2009) guidance document, damage observed in the 2013
Seddon earthquakes highlighted many design deficiencies, and there is a clear need for better wine
tank design. There appears to be a knowledge gap between the use of the NZSEE guideline and its
implementation into the wine industry. This section outlines some of the loadpaths and mechanisms,
not explicitly covered by the NZSEE document, which should be checked as part of the seismic design
of wine tanks.
665
4.1 Point bearing reaction and transition zone
For plinth-mounted tanks with ductile hold-down anchors, NZSEE (2009) allows a ductile elastic
anchor force distribution to be assumed, as shown in Figure 5. This creates a compressive point
reaction at the tip of the tank and a loadpath must be provided to resist this. The magnitude of this
point reaction, R, is given in equation 1. Further allowance will be needed for overstrength if a
capacity design approach is used.
𝑅𝑅 =
8𝑀𝑀𝑂𝑂𝑂𝑂
6𝐷𝐷
+ 𝑊𝑊𝑟𝑟𝑟𝑟
(1)
where MOT = overturning moment; D = tank diameter; and Wrw = total weight of tank roof and wall.
Vertical stress distribution in tank
wall due to overturning moment
Stress transition
zone
Ductile elastic anchor force
distribution at tank base
Compressive
point reaction, R
Figure 5. Stress distribution in tank wall and at base for tanks with ductile hold-down anchors.
If the loadpath is provided by continuing the tank skirt down to the slab, the skirt and concrete bearing
must be appropriately checked to prevent buckling of the skirt or bearing failure of the concrete slab.
If chairs are being relied upon to transfer some of this reaction, local buckling above the chair should
also be considered to avoid that seen in Figure 3f.
Further up the tank, NZSEE (2009) assumes an elastic flexural stress distribution in the tank walls, as
shown in Figure 5. It then follows that there must be a transition zone between this and the base of the
tank where the stress distribution changes to the ductile elastic anchor force distribution. It makes
sense that the increased axial compression in the tank wall above the compression reaction also be
considered.
4.2 Shear transfer to plinth
For plinth mounted tanks, shear transfer from the tank to the plinth must be considered. Typical
mechanisms include friction between the tank floor and plinth, and hoop stress within the skirt if the
plinth is constructed so that it bears against the skirt. If the tank floor is sloped, the effect of this
should be considered as part of the shear-friction mechanism, as friction resistance will reduce when
shear is directed away from the slope.
4.3 Shear transfer to slab
Similarly, shear transfer from the plinth to the slab should be considered, as often the plinth is cast
separately from the slab. Typical mechanisms include shear friction, with additional assistance from
shear dowels epoxied into the slab and cast within the plinth. Recently the authors have seen holddown anchors with baseplates, which may also provide shear resistance to the plinth.
In theory, some shear may be transferred via the compressive point reaction, similar to a reinforced
concrete shear wall, provided the skirt has sufficient out-of-plane support. Further research and
guidance would be helpful on this.
666
4.4 Hold-down anchors
In the New Zealand wine industry, tanks are often relocated to suite winery expansions or
reconfigurations. Because of this, most hold-down anchors are post-installed, as oppose to cast into the
slab with anchor plates. Wine tanks are typically arranged in lines of double-rows, with drains and
walkways between each double row.
Due to the proximity of adjacent tanks, there is interaction between the tank and foundation slab,
which means that hold-down anchors will generally be located within the tension zone of the slab
during the earthquake (Figure 6). At this location, the concrete is likely to crack which will affect the
performance of epoxied anchors. This should be considered in the design of the hold-downs, and the
slab should be appropriately designed so that expected cracks widths do not compromise the integrity
of the anchors. Overseas guidance within the last decade, such as EOTA (2007, 2013a, 2013b), have
included the design of epoxied anchors within cracked concrete.
High slab shear forces
can exist at this location
Hold-down anchors epoxied
within tension zone of slab
Figure 6. Wine tank and foundation slab interaction.
Where hold-down anchors are connected to the tank via brackets/chairs, the prying actions induced
from the eccentric connection onto the skirt should be considered. The authors have typically used
3-dimensional shell analysis to determine the expect shell bending and membrane stresses induced.
Often compensating plates welded to the skirt are required.
Wine makers generally prefer narrow slender tanks as this suits the mixing and fermentation process.
The strength of epoxy anchors generally limits wine tanks in Marlborough to a height to radius ratio of
around 4.5, when designed as an importance level 1 structure. If designed to a higher level, the tank
will generally need to be squatter.
4.5 Foundation slab
Foundation slabs need to provide adequate hold-down against overturning and limit soil bearing to suit
site conditions. In addition, any design actions from interaction between adjacent tanks should be
considered.
As illustrated in Figure 6, there can be large shears forces in the zone between adjacent tanks. From
the authors’ experience, shear reinforcement is often required for larger tanks. In addition, the
Concrete Structures Standard, NZS3101:2006, requires minimum shear reinforcement to be provided
in slabs thicker than 400mm, where the shear force exceeds half the design shear strength provided by
the concrete.
4.6 Steel base-frames
Steel base-frames should be designed for the seismic loadcase which includes both gravity and
earthquake actions simultaneously. The authors have come across existing tanks, where member sizes
are just sufficient for the gravity loads, suggesting that lateral earthquake loads had not been
considered.
667
Where frames are required to act as moment-resisting frames to provide lateral resistance, connections
need to be appropriately detailed as rigid joints. In many joints the authors have seen, CHS legs are
simply fillet welded to the underside of RHS beams. A better solution would be for RHS beams to be
butt-welded to the CHS, with horizontal stiffeners provided to the CHS which align with RHS flanges.
Where frames are braced, additional flexural actions in the legs created by eccentricities in the
alignment of braces (such as that in Figure 1a) should be considered. From the author’s assessment of
some existing frames, it is apparent this has not been done.
Failures observed in Marlborough highlight the importance of aligning straps/tags connecting the
base-frame to the tank with the tank wall and ensuring sufficient weld is provided. In addition,
threaded adjustable feet require improvement, if they are to remain in use, to ensure these are not a
weak point.
4.7 Catwalks and services
Catwalks and services attached to tanks should be connected with sliding joints such that tanks can
move laterally independently of each other without the introduction of force transfer between tanks.
Where catwalks are supported by tanks, it follows that the catwalk will have the same seismic design
level as the tank, and this should be discussed with the winery. For catwalks which are frequently
used, higher design levels than is typically used for tanks may be appropriate. Options include
supporting the catwalks on a separate structure or increasing the design level of the supporting tank.
But note the latter may require the tanks to be squatter which has further implications on the tank cost,
foundations, wine making and land use.
The foundation slab, tank, catwalks and services are often designed and installed by different parties,
and much of these issues arise from a lack of coordination between them. These issues are best dealt
with good project management at the beginning of the project, in consultation with the winery.
4.8 Roof cone
Typical roof cones of wine tanks in New Zealand do not appear to have the stiffening girders or large
radius knuckles to provide the compression rings described in API-620 (2008). As a result, crumpling
of the roof cone knuckle was observed in Marlborough following the 2013 Seddon earthquakes. To the
authors understanding, no rupture of the knuckle was observed and there was no loss of contents as a
result. Further guidance on acceptable performance levels in this area for wineries would be beneficial
for the industry.
5 FUTURE RESEARCH NEEDED
Numerical studies have shown that liquid-storage tanks with energy dissipating anchors experience a
significant reduction in base shear and overturning moment when compared to fully anchored tanks,
and exhibit smaller levels of base uplift and floor plate deformation when compared to unanchored
tanks (Malhotra 1998).
Designing tanks with energy dissipation results in numerous benefits. Firstly, cost savings will arise
from a reduction in tank wall thicknesses and foundation size. Secondly, smaller overturning moment
allows more slender tanks, which are desirable from a winemaking and land utilisation perspective.
Thirdly, reliable energy dissipation introduces a level of resilience to the system which currently does
not exist with traditional tension-only anchored tanks. When tension-only anchored tanks experience
earthquake actions larger than designed for, the result is base uplift and floor plate deformations on a
similar order to that of an unanchored tank, and base shear and overturning actions of similar
magnitude to a fully anchored tank (Malhotra and Veletsos 1995) – effectively the worst of both
worlds.
Unfortunately the design and analysis of partially uplifting tanks with energy dissipation is limited to
nonlinear methods such as that proposed in Malhotra (2000). To encourage the development and
design of such systems, further research is required to produce design guidance of similar vein to the
current NZSEE (2009) document. The authors envisage that this could be achieved via design charts
668
produced from a series of numerical studies, considering tanks of different aspect ratios, partial holddown anchorage ratios and damping.
Future research should also include investigating whether it is possible to introduce reliable capacity
design to wine tanks to limit damage to replaceable fuses. Whilst the ‘capacity design’ term is used in
the NZSEE guideline, the authors do not believe the ‘capacity design’ method described therein
provides the same protection as provided in capacity designed building structures. When a tensiononly anchored tank is overloaded, the tank will uplift, essentially increasing its hold-down capacity by
picking up the weight of the contents as the floor peels from the plinth. As such, the overturning
capacity of the tank will continue to increase as required, increasing the likelihood of buckling of the
tank wall. The opportunity to limit loads to prevent wall buckling and isolate damage to replaceable
fuses would protect the winery’s produce and allow continued operation following seismic events.
6 CONCLUSION
The NZSEE (2009) guidelines for the seismic design of liquid storage tanks provide excellent
procedures for determining design actions and analysing the performance of tanks. However given the
wide scope of the document, there is a lack of guidance that deals with the wine tank configurations
commonly used in New Zealand Wine Industry. This is apparent from observed damage to wine tanks
in Marlborough following the 2013 Seddon earthquakes.
There are loadpaths and mechanisms that need to be considered in wine tanks design, which are not
explicitly covered by the NZSEE (2009) guidelines. These include compressive point bearing
reactions from overturning, shear transfer to the plinth and slab, prying actions on the skirt where
external hold-down chairs are used, foundation shear and the influence of cracks on epoxy anchors,
adequate detailing of steel base-frame connections and effects from attached catwalks and services.
Numerical studies have shown that designing tanks with energy dissipation results in several benefits
in terms of cost, land utilisation and performance. To encourage the development and use of such
systems, further research is needed to develop guidance for tanks with energy dissipation on similar
vein to the existing NZSEE guidance.
7 REFERENCES
American Petroleum Institute (API). (11th ed.) 2008. Design and Construction of Large, Welded, Low-pressure
Storage Tanks, API Standard 620, USA: API.
EOTA. 2007. Design of Bonded Anchors, Technical Report TR-029, European Organisation for Technical
Approvals.
EOTA. 2013a. Design of Metal Anchors for use in Concrete under Seismic Actions, Technical Report TR-45,
European Organisation for Technical Approvals.
EOTA. 2013b. ETAG 001 Metal Anchors for Use in Concrete, European Technical Approval Guidelines
(ETAG) 001.
Malhotra, P.K. 1998. Seismic Strengthening of Liquid-Storage Tanks with Energy-Dissipating Anchors, Journal
of Structural Engineering, Vol 124(4) 405-414 American Society of Civil Engineers (ASCE).
Malhotra, P.K. 2000. Practical Nonlinear Seismic Analysis of Tanks, Earthquake Spectra, Vol 16(2) 473-492
Earthquake Engineering Research Institute (EERI).
Malhotra, P.K. & Veletsos, A.S. 1995. Seismic Response of Unanchored and Partially Anchored Liquid-Storage
Tanks, Electric Power Research Institute (EPRI) TR-105809.
Morris, G.J., Bradley, B.A., Walker, A.F. & Matuschka, T. 2013. Ground Motions and Damage Observations in
the Marlborough Region from the 2013 Lake Grassmere Earthquake, Bulletin of the New Zealand Society for
Earthquake Engineering, Vol 46(4) 169-187New Zealand Society of Earthquake Engineering (NZSEE),
2009. Seismic Design of Storage Tanks: 2009. Wellington: NZSEE
NZIER. 2009. Economic Impact of the New Zealand Wine Industry. An NZIER report to New Zealand
Winegrowers.
Standards New Zealand. 1997. Steel Structures Standard. NZS3404:1997, Wellington, New Zealand.
669
Standards New Zealand. 2004. Structural Design Actions, Part 5: Earthquake Actions – New Zealand.
AS/NZS1170.5:2004, Wellington, New Zealand.
Standards New Zealand. 2006. Concrete Structures Standard. NZS3101:2006, Wellington, New Zealand.
670